The NIH Roadmap Epigenomics project has generated substantial genome-wide histone modification maps in a large number of human normal cells and tissues using the chromatin immunoprecipitation and sequencing (ChIP-Seq) approach. The comparison of these """"""""reference"""""""" epigenomic profiles with those in human cancer cells (e.g., from ENCODE) will greatly advance our understanding of the mechanisms of cancer initiation and progression and will guide therapeutic studies on human cancers. However, there are substantial analytic challenges for such comparisons: (1) the """"""""peaks"""""""" called from the ChIP-Seq data are usually broad regions, ranging from several hundred to several million base pairs for a single histone marker;(2) the peaks from ChIP-Seq are limited in resolution due to the sonication process, so the precise locations and sizes of the peaks for the same marker vary from experiment to experiment;(3) the cross-linking protocol by formaldehyde introduces noise or non-specific signals of histone markers. We have previously demonstrated that histone modification at the resolution of the single nucleosome is critical for the comparison of epigenomic profiles between different cells or tissues. Here, we propose a genomic and computational approach to decompose the ChIP-Seq signals to the resolution of individual nucleosomes and then compare histone modifications of each nucleosome between samples. Given that nucleosome mapping is not the focus for either the Roadmap Epigenomics project or the ENCODE project, we will generate high-resolution nucleosome maps for human normal and cancer cells and incorporate these nucleosome data and also histone modification data at the resolution of the single nucleosome into both projects by building a nucleosome genome browser.
Our proposed study aims to develop a novel algorithm for the identification of differential histone modification at the resolution of the single nucleosome between stem and differentiated cells and between normal and cancer cells, respectively. We will use nucleosome mapping data in these cells and facilitate the comparison of histone modifications between different cells.